Illustrative report preview
The verdict
PromisingMixedContradicted
- DemandGrounded 12 sample sources
- CompetitorsGrounded 9 sample sources
- PricingGrounded 6 sample sources
- Build firstInferred Interpretation
- Prove nextInferred Open question
Agentic market research & product validation
Research and validate your startup idea at the speed of AI.
Describe your idea. Clien researches the market, competitors, customers, communities and pricing, then gives you an evidence-backed brief on what to build, what to test, and what to do next.
The problem
The product works, but no one cares.
AI helps you build faster. It doesn’t make the problem worth solving, the product worth switching to, or the customer willing to pay.
Building feels like progress. Research feels like friction. So the biggest assumptions get tested after they’re baked into the product. Customer feedback should test an informed hypothesis—not rescue an uninformed guess.
Why this keeps happeningOur coding process is living in 2026, but our market research process is still stuck in 2010.
Lily, Indie Hackers · indiehackers.com ↗
How it works
Agents research your idea. You make the informed decision.
Your ideaOne research question
- ForumsGrounded 14 threads
- Open webGrounded 22 pages read
- CompetitorsGrounded 7 profiles
- InterviewsSynthetic 3 transcripts
- MarketInferred TAM · SAM · SOM
One research brief PromisingDo · Don't · Watch
What you getOne research brief.
One research brief.
A much clearer next move.
Market, competitors, customers, pricing, risks, what to build first, and what to test next—researched, sourced and synthesised for you.
Answers to your questions
- 01
Is this worth pursuing?
Demand signals, customer pain, competitors, market gaps and pricing benchmarks.
- 02
What should I build?
The strongest target customer, positioning, MVP scope and feature priorities.
- 03
What do I need to prove?
Ranked assumptions, biggest risks, evidence gaps and questions to take to real customers.
Not just one report
Keep researching your idea as it changes.
A report is the start. Return when you're choosing a feature, testing a price, or considering a pivot. New research can build on the project instead of starting from a blank page.
Keep reports, competitor scans, community signals and the questions still worth testing in one research trail.
Research whether an async stand-up tool for remote teams has real demand.
Clien tools used
- Completed
get_credit_balanceCheck the available balance before planning paid work - Completed
clien_researchValidate the product idea end to end - Completed
get_reportRe-open the evidence-backed report without a new run
Illustrative AI response
The research points to recurring coordination pain, but the willingness-to-pay assumption still needs a direct test. I've kept the evidence and open questions with this project.
Clien MCP
Bring Clien into your AI assistant.
Clien runs as a local MCP server in compatible assistants such as Claude Code and Codex. Ask for a full validation, a focused competitor or community scan, or a read of research the project already holds—and, with your authorisation, start new research without leaving the conversation.
The one-command installer currently configures Claude Code automatically. Other compatible MCP clients require manual setup.
npx @clien-ai/mcp installRead the install docs (opens in a new tab)Pricing
If your first report doesn’t give you one useful, cited insight for your next product decision, get your entire first Pro charge back.
One report
$12one-time
- Fully research and validate one idea
- One full report
- No subscription
Pro
Best value$49per month
Equivalent to $41 per month
- 300 credits every month
Typical research costs
Typical research costs- Full research report
- 50 cr
- Market research
- 5 cr
- AI interview
- 3 cr
- Competitor research
- 3 cr
- Report synthesis
- 3 cr
- Community research
- 2 cr
- Persona generation
- 1 cr
- Persona enrichment
- 1 cr
- Up to 6 full reports
- Full and focused research
- Unlimited projects
- Top up anytime
- Cancel anytime
FAQ
Fair questions
Straight answers on what Clien can and can't do for you.
Will Clien tell me whether my idea will succeed?
No, and be sceptical of anything that claims it can. Clien assembles what is already knowable—who else is doing this, what people with the problem say about it, and what comparable products charge—and tells you which assumptions look risky. That sharpens your judgment and points at what to test. It doesn't predict the future.
How is this different from asking ChatGPT or Claude?
ChatGPT and Claude can do strong research, especially with a careful prompt. Clien makes the process repeatable and auditable: it researches the market, competitors and public customer discussions in separate passes, tests explicit hypotheses, and saves the result as a report where claims are linked to sources or clearly marked as inference or unsupported. It also shows what the research could not establish. The advantage isn’t magic intelligence; it’s a dedicated research workflow and an evidence trail you can inspect and reuse.
Does this replace talking to customers?
No. It makes those conversations better aimed. By the time you speak to someone, you already know what's been asked and answered elsewhere, so you can spend the call on what's genuinely unresolved. Every report ends with the questions it couldn't settle, written as something you can actually ask.
What exactly does Clien research?
A full report researches the market and category, competitors and their positioning, public discussions from people with the problem, and the assumptions your idea depends on. Clien also interviews synthetic users built from that public evidence and labels them clearly as AI rather than real respondents.
What can I investigate after the first report?
You can run another full pass when the idea has genuinely moved, or investigate a narrower question with focused competitor and community research. Those results land in the same project, so your evidence builds over time instead of starting again.
How do credits work?
Credits pay for new AI research. A full research report costs 50 credits; focused research costs less—for example, market research costs 5, competitor research costs 3, community research costs 2.
Pro includes 300 credits every month, enough for up to 6 full reports. You can top up when you need more.
Reading research already in your projects does not use credits. Technical failures that produce no usable output restore the affected credits automatically.
Can I use my research inside Claude or another assistant?
Yes. Clien runs as an MCP server, so you can use it from compatible assistants such as Claude Code or Codex. The one-command installer currently configures Claude Code; other MCP clients require manual setup. Once connected, your assistant can read your Clien projects and reports and—with your authorisation—start new research.
Is my idea private?
Yes. Your projects and reports are private by default and protected by account access controls. Clien uses service providers to host the product and perform research.
Clien sends relevant parts of your idea to third-party AI and search providers to perform the research. If you deliberately create a share link, anyone with that link can view the shared report, and you can revoke it at any time.
Read our Privacy Policy
Who's building Clien
Built by someone who's made the same decisions.
I'm Alex Omeyer. I built and exited a VC-backed startup, work in product management, and have been building AI products since 2015.
AI made building fast. It didn't make choosing what to build any easier. That's why I built Clien.